Slanted Images: Measuring Nonverbal Media Bias During the 2016 Election

43 Pages Posted: 6 May 2021

Date Written: April 15, 2021

Abstract

Using nearly one million images from the front page of news websites during the 2016 election period, I show how computer vision techniques can identify the faces of politicians across the images and measure the nonverbal emotional content expressed on each face. I find strong evidence for nonverbal media bias in both the choice of which politicians to cover and the emotional content of the images used. Liberal websites devoted 40 (14) percent of their visual political coverage to Donald Trump (Hillary Clinton) compared to 30 (25) percent among conservative outlets. Websites whose consumers are politically aligned with a candidate also portray the candidate with more positive emotions and less negative emotions than non-aligned websites. Moreover, I find evidence for important dynamics across the election cycle, with the partisan gap in who to cover increasing significantly after the primaries.

Keywords: media bias, images as data, elections, machine learning, emotions

JEL Classification: P16, D70, D80, D22

Suggested Citation

Boxell, Levi, Slanted Images: Measuring Nonverbal Media Bias During the 2016 Election (April 15, 2021). Available at SSRN: https://ssrn.com/abstract=3837521 or http://dx.doi.org/10.2139/ssrn.3837521

Levi Boxell (Contact Author)

Stanford University ( email )

Stanford, CA
United States

Do you have a job opening that you would like to promote on SSRN?

Paper statistics

Downloads
11
Abstract Views
102
PlumX Metrics